Charles W. Fox

A tutorial on variational Bayesian inference

This tutorial describes the mean-field variational Bayesian approximation to inference in graphical models, using modern machine learning terminology rather than statistical physics concepts. It begins by seeking to find an approximate mean-field distribution close to the target joint in the KL-divergence sense. It then derives local node updates and reviews the recent Variational Message Passing framework.

Tactile Discrimination Using Active Whisker Sensors

We describe a novel, biomimetic tactile sensing system modeled on the facial whiskers (vibrissae) of animals such as rats and mice. The “BIOTACT Sensor” consists of a conical array of modular, actuated hair-like elements, each instrumented at the base to accurately detect deflections of the shaft by whisker-surface contacts. A notable characteristic of this array is that, like the biological sensory system it mimics, the whiskers are moved back-and-forth (“whisked”) so as to make repeated, brief contacts with surfaces of interest. Furthermore, these movements are feedback-modulated in a manner intended to emulate some of the “active sensing” control strategies observed in whiskered animals. We show that accurate classification of surface texture using data obtained from whisking against three different surfaces is achievable using classifiers based on either naive Bayes or template methods. Notably, the performance of both these approaches to classify textures after training on as few as one or two surface contacts was improved when the whisking motion was controlled using a sensory feedback mechanism. We conclude that active vibrissal sensing could likewise be a useful sensory capacity for autonomous robots.

Contact type dependency of texture classification in a whiskered mobile robot

Actuated artificial whiskers modeled on rat macrovibrissae can provide effective tactile sensor systems for autonomous robots. This article focuses on texture classification using artificial whiskers and addresses a limitation of previous studies, namely, their use of whisker deflection signals obtained under relatively constrained experimental conditions. Here we consider the classification of signals obtained from a whiskered robot required to explore different surface textures from a range of orientations and distances. This procedure resulted in a variety of deflection signals for any given texture. Using a standard Gaussian classifier we show, using both hand-picked features and ones derived from studies of rat vibrissal processing, that a robust rough-smooth discrimination is achievable without any knowledge of how the whisker interacts with the investigated object. On the other hand, finer discriminations appear to require knowledge of the target’s relative position and/or of the manner in which the whisker contact its surface.

Naive Bayes texture classification applied to whisker data from a moving robot

Many rodents use their whiskers to distinguish objects by surface texture. To examine possible mechanisms for this discrimination, data from an artificial whisker attached to a moving robot was used to test texture classification algorithms. This data was examined previously using a template-based classifier of the whisker vibration power spectrum [1]. Motivated by a proposal about the neural computations underlying sensory decision making [2], we classified the raw whisker signal using the related ‘naive Bayes’ method. The integration time window is important, with roughly 100ms of data required for good decisions and 500ms for the best decisions. For stereotyped motion, the classifier achieved hit rates of about 80% using a single (horizontal or vertical) stream of vibration data and 90% using both streams. Similar hit rates were achieved on natural data, apart from a single case in which the performance was only about 55%. Therefore this application of naive Bayes represents a biologically motivated algorithm that can perform well in a real-world robot task.

Optimal decision-making in mammals: insights from a robot study of rodent texture discrimination

Texture perception is studied here in a physical model of the rat whisker system consisting of a robot equipped with a biomimetic vibrissal sensor. Investigations of whisker motion in rodents have led to several explanations for texture discrimination, such as resonance or stick-slips. Meanwhile, electrophysiological studies of decision-making in monkeys have suggested a neural mechanism of evidence accumulation to threshold for competing percepts, described by a probabilistic model of Bayesian sequential analysis. For our robot whisker data, we find that variable reaction-time decision-making with sequential analysis performs better than the fixed response-time maximum-likelihood estimation. These probabilistic classifiers also use whatever available features of the whisker signals aid the discrimination, giving improved performance over a single-feature strategy, such as matching the peak power spectra of whisker vibrations. These results cast new light on how the various proposals for texture discrimination in rodents depend on the whisker contact mechanics and suggest the possibility of a common account of decision-making across mammalian species.

Tactile SLAM with a biomimetic whiskered robot

Future robots may need to navigate where visual sensors fail. Touch sensors provide an alternative modality, largely unexplored in the context of robotic map building. We present the first results in grid based simultaneous localisation and mapping (SLAM) with biomimetic whisker sensors, and show how multi-whisker features coupled with priors about straight edges in the world can boost its performance. Our results are from a simple, small environment but are intended as a first baseline to measure future algorithms against.

Whisker-object contact speed affects radial distance estimation

Whiskered mammals such as rats are experts in tactile perception. By actively palpating surfaces with their whiskers, rats and mice are capable of acute texture discrimination and shape perception. We present a novel system for investigating whisker-object contacts repeatably and reliably. Using an XY positioning robot and a biomimetic artificial whisker we can generate signals for different whisker-object contacts under a wide range of conditions. Our system is also capable of dynamically altering the velocity and direction of the contact based on sensory signals. This provides a means for investigating sensory motor interaction in the tactile domain. Here we implement active contact control, and investigate the effect that speed has on radial distance estimation when using different features for classification. In the case of a moving object contacting a whisker, magnitude of deflection can be ambiguous in distinguishing a nearby object moving slowly from a more distant object moving rapidly. This ambiguity can be resolved by finding robust features for contact speed, which then informs classification of radial distance. Our results are verified on a dataset from SCRATCHbot, a whiskered mobile robot. Building whiskered robots and modelling these tactile perception capabilities would allow exploration and navigation in environments where other sensory modalities are impaired, for example in dark, dusty or loud environments such as disaster areas.

Hippocampus as unitary coherent particle filter

We present a mapping of the hippocampal formation onto a Temporal Restricted Boltzmann Machine [1] based architecture, running a deterministic version of Gibbs sampling, and extended with a lostness detection and recovery circuit modelled on subiculum and septal acetylcholine (ACh). The mapping approximates Bayesian filtering, which infers both auto-associative de-noised percepts and temporal sequences, the latter including sequences of places during navigation. Inference may be viewed as a neurally implemented particle filter with a single particle - as suggested previously [2] as a purely behavioural animal model.

Simultaneous localisation and mapping on a multi-degree of freedom biomimetic whiskered robot

A biomimetic mobile robot called “Shrewbot” has been built as part of a neuroethological study of the mammalian facial whisker sensory system. This platform has been used to further evaluate the problem space of whisker based tactile Simultaneous Localisation And Mapping (tSLAM). Shrewbot uses a biomorphic 3-dimensional array of active whiskers and a model of action selection based on tactile sensory attention to explore a circular walled arena sparsely populated with simple geometric shapes. Datasets taken during this exploration have been used to parameterise an approach to localisation and mapping based on probabilistic occupancy grids. We present the results of this work and conclude that simultaneous localisation and mapping is possible given only noisy odometry and tactile information from a 3-dimensional array of active biomimetic whiskers and no prior information of features in the environment.

Naive Bayes novelty detection for a moving robot with whiskers

Novelty detection would be a useful ability for any autonomous robot that seeks to categorize a new environment or notice unexpected changes in its present one. A biomimetic robot (SCRATCHbot) inspired by the rat whisker system was here used to examine the performance of a novelty detection algorithm based on a “naive” implementation of Bayes rule. Naive Bayes algorithms are known to be both efficient and effective, and also have links with proposed neural mechanisms for decision making. To examine novelty detection, the robot first used its whiskers to sense an empty floor, after which it was tested with a textured strip placed in its path. Given only its experience of the familiar situation, the robot was able to distinguish the novel event and localize it in time. Performance increased with the number of whiskers, indicating benefits from integrating over multiple streams of information. Considering the generality of the algorithm, we suggest that such novelty detection could have widespread applicability as a trigger to react to important features in the robots environment.